Download Vintage Capital and Creditor Protection Please share

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Transcript
Vintage Capital and Creditor Protection
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Benmelech, Efraim and Nittai K. Bergman. "Vintage Capital and
Creditor Protection." NBER Working Papers, Feb. 2010,
no.15735 (Working papers series (National Bureau of Economic
Research)) http://www.nber.org/papers/w15735.pdf
As Published
http://ideas.repec.org/p/nbr/nberwo/15735.html
Publisher
National Bureau of Economic Research
Version
Author's final manuscript
Accessed
Thu May 26 23:36:28 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/64451
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike 3.0
Detailed Terms
http://creativecommons.org/licenses/by-nc-sa/3.0/
Vintage Capital and Creditor Protection∗
Efraim Benmelech
Harvard University and NBER
Nittai K. Bergman
MIT Sloan and NBER
∗
We thank Marios Angeletos, Douglas Baird, Lucian Bebchuk, Guido Imbens, Boyan Jovanovic, Florencio Lopez-deSilanes, Giacomo Ponzetto, Adriano Rampini, Andrei Shleifer, Jeremy Stein, and seminar participants at Berkeley,
Columbia Law School, Duke, the European Summer Symposium in Financial Markets, Federal Reserve Bank of
Boston, Harvard Economics, Harvard Law School, London Business School, London School of Economics, Northwestern, Ohio State University, Stanford, Stockholm School of Economics, Tilburg University, 2008 WFA meetings
in Waikoloa, University of Alberta, University of Amsterdam, and the University of Illinois at Urbana-Champaign,
and Yale Law School. We also thank Robert Grundy and Phil Shewring from Airclaims Inc. Alex Radu and Kate
Waldock provided excellent research assistance. All errors are our own.
Efraim Benmelech, Department of Economics, Harvard University, Littauer Center, Cambridge, MA 02138. E-mail:
effi benmelech@harvard.edu. Nittai Bergman, Sloan School of Management, MIT, 50 Memorial Drive, Cambridge,
MA 02142. E-mail: nbergman@mit.edu.
Vintage Capital and Creditor Protection
Abstract
We provide novel evidence linking the level of creditor protection provided by law to the degree of
usage of technologically older, vintage capital in the airline industry. Using a panel of aircraft-level
data around the world, we find that better creditor rights are associated with both aircraft of a
younger vintage and newer technology as well as firms with larger aircraft fleets. We propose that
by mitigating financial shortfalls, enhanced legal protection of creditors facilitates the ability of
firms to make large capital investments, adapt advanced technologies and foster productivity.
Introduction
There is a large body of evidence that better legal rules covering protection of corporate shareholders
and creditors are associated with more developed financial markets and higher economic growth (La
Porta et al., (1997), (1998); King and Levine, (1993); Beck et. al., (2000); and Rajan and Zingales,
(1998)). While the empirical regularities found in the data are quite robust, most of the research
is based on cross-country outcomes and suffers from small samples and potential identification
problems (see Djankov at al. (2007)). In addition, the results from cross-country regressions do
not pin down the underlying mechanism through which creditor rights and shareholder protection
affect real economic outcomes. This paper attempts to fill this gap. We study the relation between
creditor protection and the use of vintage capital in the airline industry in a sample of most of the
aircraft in the world (489,916 aircraft-year observations) covering 5,987 operators in 129 countries
in the years 1978-2003. We find that airlines enjoying the benefits of higher creditor protection
operate aircraft of a newer technology and younger vintage.
The importance of new capital goods for economic growth has been suggested by Solow (1960):
“...many if not most innovations need to be embodied in new kinds of durable equipment before
they can be made effective. Improvements in technology affect output only to the extent that
they are carried into practice either by net capital formation or by the replacement of old-fashioned
equipment by the latest models...” More recent theoretical models show that capital of older vintage
hampers productivity and growth (Benhabib and Rustichini (1991), Hsieh (2001)), slows technology diffusion (Chari and Hopenhayn (1991)), and increases income inequality across individuals
and countries (Jovanovic (1998)). Empirical estimates suggest that around 60% of US per-capita
growth is due to technical change that is embodied in new more efficient capital goods (Greenwood, Hercowitz, and Krusell (1997)).1 Our paper provides novel evidence on a financial channel
in technological adaption and capital formation.
While we propose and provide evidence on one mechanism connecting financial constraints
and creditor protection to aircraft vintage and fleet size, our results suggest a broader link between
financial development, investor protection and economic activity. Our empirical methodology differs
from previous research which has focused mostly on aggregate, macroeconomic outcomes of investor
protection such as financial market development and economic growth (King and Levine (1993), La
1
See Boucekkine, de la Croix, and Licandro (2008) for a survey of the vintage capital literature.
1
Porta et al. (1997, 1998), Rajan and Zingales (1998)). The wealth of the data and our focus on an
important global industry allow careful consideration and identification of the specific mechanism
through which investor protection affects and fosters technical progress and economic development.2
We start by developing a simple price-theory model of an airline choosing its scale and average
asset age given an internal financing constraint and an external creditor protection environment
that is determined at the country level. The airline must decide on the quantity of aircraft to
purchase and their average age. Older aircrafts are assumed to be less efficient – either because
of depreciation in aircraft efficiency stemming from their normal use, or because of technological
improvements in aircraft design over time. The model shows that increased availability of external
finance due to enhanced creditor protection will have two important effects on firms. First, when
creditor rights are greater, and hence financial constraints more relaxed, firms will be able to invest
in newer, more expensive technologies. Second, since financing considerations will place fewer
constraints on firm scale, firms will tend to be larger in countries with greater creditor rights.3
Using detailed profiles of most aircraft in the world during the period 1978-2003 we then study
the relation between the level of creditor protection and two measures of aircraft vintage. The first
measure of vintage is simply aircraft age, defined as the time elapsed since the date of the aircraft
delivery. The second measure of vintage, called ‘technological age’, is calculated as the time elapsed
since the model type of that aircraft was first introduced. The level of country creditor protection
is measured using the creditor rights score as developed by La Porta et al. (1997, 1998), and in
particular the more recent score that covers 129 countries in the years 1978-2003 (Djankov et al.
(2007)). Consistent with our first prediction, our analysis shows that aircraft operated in countries
with higher creditor protection are of a younger vintage and newer technology. Furthermore, we
also find that operators’ size are larger in countries with better creditor protection. Our regressions
controls for a battery of economic development variables, legal origin, government ownership of
airlines, and a country’s civil aviation quality, and a variety of year, country, and airline fixed2
Our paper adds to a growing body of literature that uses industry- and firm-level data to evaluate the effects of
investor protection and financial development on resource allocation (Fisman and Love (2004), Wurgler (2000)), economic growth (Demirguc-Kunt and Maksimovic (1998), Guiso, Sapienza, and Zingales (2005)), corporate risk-taking
and innovation (Acharya, Amihud, and Litov (2008), Acharya and Subramanian (2008)), and financial contracts and
lending structures (Bergman and Nicolaievsky (2007), Braun (2003), Esty and Megginson (2003), Lerner and Schoar
(2005), Liberty and Mian (2007), Onega and Smith (2000), and Qian and Strahan (2007)).
3
Our model is closely related to Eisfeldt and Rampini (2007) who show that firms which are credit constrained
purchase more used, rather than new, capital because, higher ex-post maintenance payments of used capital relaxes
current ex-ante financial constraints.
2
effects. In particular, the panel dimension of our data allows us to control for country fixed-effects
and hence to identify off of changes in creditor rights within a country.
To alleviate concerns about omitted variables and to provide additional evidence in support
of the financing channel of technology adoption, we conduct a number of tests which split our
sample into aircraft that should be treated by the creditor rights index and those that should not
be treated by this index. We begin by splitting our sample into aircraft operated by commercial
and private airlines, and those operated by the military. We expect the negative relation between
creditor protection and both aircraft age and fleet size to hold only for non-military operators,
since private and commercial operators are those required to raise funds from outside investors in
cases of cash flow shortages. Moreover, only commercial and private operators would fall under
the bankruptcy provisions of the local corporate and bankruptcy laws which are the essence of the
creditor rights score. In contrast, sovereign debtors are incentivized to repay creditors mainly for
reputational concerns and continued access to capital markets (Bulow and Rogoff (1989a,b)). Our
results confirm this first conjecture: we find that the creditor rights score is correlated with the
age and fleet size of commercial and private operators but is uncorrelated with the age and size of
military fleets.
Focusing on commercial and private operators, we then split our sample into planes that are
leased and those which are not leased. Following Eisfeldt and Rampini (2008), we conjecture that
leasing allows firms to alleviate some of the financial frictions associated with debt financing, as asset
repossession is easier for a lessor than for a creditor. This difference in financing frictions implies,
first, that airlines in countries with poor creditor rights will be more likely to lease rather than
own aircraft, and second, that the negative relation between creditor rights and aircraft vintage
described above should be concentrated amongst non-leased aircraft. We find support for both
of these hypotheses in the data: i) aircraft are more likely to be leased in countries with worse
creditor protection, and ii) while there is a strong statistically significant negative relation between
creditor rights and non-leased aircraft, we find no relation between creditor rights and aircraft
vintage amongst leased aircraft. By examining the relation between creditor right and both leased
and non-leased aircraft separately, we alleviate the concern that our results are driven by variation
in unobserved variables, and in particular variation in investment opportunities, correlated with
variation in creditor rights. Indeed, there is little reason to suspect that increased investment
opportunities should differentially impact the vintage of leased vs. non-leased aircraft. In contrast,
3
the financing channel provides a clear prediction regarding the differential impact of creditor rights
on the two methods of aircraft financing.
We continue by splitting the sample of commercial aircraft based on airlines’ financial condition.
The financing channel predicts that the negative relation between aircraft vintage and creditor
rights should be stronger for airlines in poor financial health. We find that vintage of aircraft of
airlines with lower leverage ratios and airlines with less debt overhang is less sensitive to creditor
rights. While both leverage and long-term debt are clearly endogenous, our identification strategy
relies on the interaction effects between country and firm characteristics. Furthermore, testing this
prediction also alleviates the concern that a correlation between creditor rights and unobserved
investment opportunities is driving our results, since there is little reason to suspect that increases
in creditor rights are more strongly correlated with improved investment opportunities in financially
constrained firms as compared to financially unconstrained firms.
The rest of the paper is organized as follows. Section I presents a simple price-theory model.
Section II provides a description of our data sources and summary statistics. Section III presents
the empirical link between aircraft age and utilization and efficiency. Section IV describes the
empirical analysis of the relation between creditor rights and aircraft age and fleet size. Section V
concludes.
I.
The Model
We begin by providing a simple model of a firm facing an external financing constraint which needs
to choose the vintage of the technology it will operate and its scale of operation. For simplicity,
firms in our model will choose between two technologies – new and old. Our main goal is to describe
the cross sectional variation in the allocation of vintage capital across firms operating in countries
with creditor protection, and hence financial constraints, of varying degree. The model is related to
Eisfeldt and Rampini (2007b), but assumes that technologies of different vintage are characterized
by different production functions.
A.
Technology allocation with exogenous prices
Consider a continuous set of firms deciding on their scale of operation and deciding between the
use of assets which embody either an old or a new technology. For consistency with the empirical
section we refer to firms as airlines and their assets as aircraft. For simplicity, we assume that
4
airlines can use only one type of technology in their fleets.
A fleet of qnew new aircraft is assumed to provide revenue f (qnew ), where f is twice differentiable,
concave, and f (0) = 0. Similarly, a fleet of qold old aircraft is assumed to provide revenue g(qold),
where again, g is twice differentiable, concave, and g(0) = 0. As is common, concavity of the
production function stems from decreasing returns to scale.
Airlines are assumed to be price takers. New technology aircraft are supplied perfectly elastically
at a price normalized to one while the price of an old aircraft is assumed in this section to be
exogenously given at pold .
An important aspect in the model will be the tradeoff between airline size and fleet quality.
Given a fixed amount of capital expenditure, an airline will need to choose between operating a
relatively large number of old technology aircraft, or a smaller number of new technology aircraft.
To solve this problem it turns out useful to define an equivalence function between old and new
aircraft, h, which relates a fleet of new aircraft of size qnew to the size of the old aircraft fleet with
equal revenue. Formally, h satisfies g(h(qnew )) = f (qnew ), so that h(qnew ) old aircraft provide equal
revenue as qnew new aircraft.4
Initially, we assume that airlines have no internal funds and must purchase their fleets using
funds raised in an external capital market. Each airline operates in a country with a level of
protection provided to investors parameterized by μ, where μ measures the fraction of revenue that
insiders within the airline can pledge to outside investors.5 Thus, given any revenue R, the airline’s
pledgeable income – i.e. the maximal amount that it can guarantee as repayment to its investors
– is μ ∗ R, with μ between zero and one. Capital markets are assumed to be perfectly competitive,
and the discount factor is taken for simplicity to be 1.
To choose the size and technology of its fleet, each airline will compare the value of a new fleet
to the value of an old one. The value of a fleet comprised of new aircraft in a country with investor
protection μ is given by:
Vnew (μ) = M ax[f (qnew ) − qnew ]
(1)
s.t. qnew ≤ μf (qnew )
4
Since new aircraft are assumed to be more efficient than older aircraft – either because of depreciation in aircraft
efficiency stemming from their normal use, or because of technological improvements in aircraft design over time –
we have that h(qnew ) > qnew for all qnew .
5
1 − µ can be interpreted as the fraction of revenue insiders can costlessly expropriate from outside investors.
5
Similarly, the value of a fleet comprised of old aircraft in a country with investor protection μ is
given by:
Vold (μ, pold) = M ax[g(qold ) − pold qold ]
(2)
s.t. pold qold ≤ μg(qold )
Since capital markets are perfectly competitive, outside investors break even, so that the maximand
of the maximization problems have airlines obtaining the full NPV of the project subject to the
financing constraint. We further assume that Vnew (1) > Vold (1, pold) so that in an economy without
financial constraints (or in one in which firms can pledge all of their output to investors), the new
technology is superior to the old technology.6
It is easy to see that the solution to (1), the new technology maximization problem, is given
uc to be the solution
in the following manner (the old technology solution is analogous). Define qnew
uc
satisfies the financing constraint
to the unconstrained problem, i.e. M ax(q) (f (q) − q). If qnew
q uc ≤ μf (q uc ), and is hence achievable by the airline, it will be the constrained solution as well.
On the other hand, if q uc is not achievable (q uc > μf (q uc )), the financing constraint will be binding
and the constrained solution will be defined implicitly by q = μf (q). In this region, investment and
firm value will be increasing in μ, as increases in μ relax the financing constraint.
In choosing between the new and old technologies, an airline in a country with investor protection μ simply compares Vnew (μ) to Vold(μ, p). Proposition 1 describes this choice (all proofs are
provided in the appendix):
Proposition 1. If h, the equivalency function between new and old technology, is convex in qnew ,
then for every pold there exists a μ̄ ≥ 0 such that airlines with μ < μ̄ choose the old technology,
and airlines with μ > μ̄ choose the new technology.
Proposition 1 states that if h is convex, the allocation of vintage capital is such that airlines
operating in low investor protection countries will choose an old aircraft fleet, while those operating
in high investor protection countries will choose a new aircraft fleet.
To understand Proposition 1, it is useful to combine maximization problems (1) and (2) into
one, by realizing that, in effect, each airline can produce revenue f (qnew ) in two ways – either
by employing qnew new aircraft or by employing h(qnew ) old aircraft. Thus, an airline can obtain
6
This assumption is for expositional use only, and is no longer required once pold is endogenized.
6
revenue f (qnew ) at an effective cost of c(qnew ) = min[qnew , pold ∗ h(qnew )]. The maximization
problem of an airline in a country with investor protection μ can be written therefore as
M ax[f (qnew ) − c(qnew )]
(3)
s.t. c(qnew ) ≤ μf (qnew )
Figure 1 depicts maximization problem 3 graphically in the case where h is convex in qnew . As
can be seen, the convexity of h is equivalent to stating that the old technology has a comparative
advantage when a firm operates on a small scale, while the new technology has a comparative
advantage when the firm operates on a large scale. Indeed, if an airline operates in the region
where revenue f (qnew ) is comparatively low, then pold ∗ h(qnew ) < qnew – i.e. the old technology is
the efficient method of production. In contrast, if an airline operates in the region where revenue
f (qnew ) is comparatively high then pold ∗ h(qnew ) > qnew and the new technology represents the
efficient method of production. Economically, convexity of h arises if employing new technology
aircraft economizes on firm scale and organizational costs are convex in scale.
To understand Proposition 1 note that when an airline operates in a low investor protection
country (i.e. low μ) its financial constraint will be binding and thus its operating scale and revenue
will be limited. It will therefore choose to operate the old technology because in low scale production
this is the efficient method of production. As the degree of investor protection improves the financial
constraint relaxes, and firm operating scale increases. Firms therefore switch to the new technology
since it is the more efficient method of production at higher levels of production.
Figure 2 presents the value functions of the two technologies as a function of μ. As Proposition
1 states, low μ firms select the old technology, while high μ firms select the new technology.7
B.
Endogenous prices and fleet size
We now endogenize the price of old technology, pold . To do so, we assume that there is a finite
measure of preexisting old aircraft and that μ is distributed according to some distribution function
G. An equilibrium p∗old is a price for old technology aircraft such that the market for old aircraft
clears. The equilibrium is characterized by the following proposition:
Proposition 2. If h, the equivalency function between new and old technology, is convex in qnew ,
then the equilibrium p∗old is such that there exists a μ̄ > 0 such that airlines with μ < μ̄ choose the
7
Both value functions become flat when the financing constraint becomes non-binding
7
old technology, and airlines with μ > μ̄ choose the new technology.
The intuition behind Proposition 2 is straightforward. All else equal, as the price of old technology, pold , decreases, Vold(μ, pold ) increases compared to Vnew (μ). Old technology therefore becomes
more attractive and a larger fraction of airlines select it. The price of old technology simply decreases to the point where the demand for old technology equals the supply. Further, from Proposition 1, we know that at the equilibrium price p∗old it is the low μ firms which are the ones who
choose the old technology while the high μ firms choose the new technology. We next characterize
the size of airlines’ fleets as measured by their number of aircraft.
Proposition 3. There exist μ∗ and μ∗∗ such that fleet sizes of airlines with μ ≤ μ∗ are smaller
than the fleet sizes of airlines with μ ≥ μ∗∗
Proposition 3 states that airlines operating in low μ countries are financially constrained, so
that their fleet sizes are restricted. In contrast, those operating in relatively high μ countries will
not be constrained, and indeed their fleet sizes will be equal to the unconstrained level.
From Propositions 2 and 3 we have the following two predictions which are tested in the empirical
section:
Prediction 1: All else equal, airlines operating in countries with lower investor protection will
have older vintage fleets.
Prediction 2: All else equal, airlines operating in low investor protection countries will have
smaller fleets than those operating in high investor protection countries.
II.
Data and Summary Statistics
This section describes the data sources used in the empirical analysis and presents summary statistics for both aircraft age and fleet size.
A.
Aircraft Level Data
Throughout our analysis we utilize data from the Ascend CASE database – a leading provider of
individual aircraft and airline data which contains ownership and operating information about most
commercial and corporate aircraft worldwide as well as many military and government aircraft. We
construct a sample of all aircraft that are available in the database for the 129 countries that are
included in Djankov at al. (2007). Our sample consists of all aircraft worldwide over the period
8
January 1, 1978 to December 31, 2003 in the Ascend CASE database.8 The data are very detailed
and include information on individual aircraft characteristics such as model-type, serial number,
year of construction, operating airline, and owner. The data in Ascend CASE thus enables us to
uniquely identify most of the aircraft in the world during the time period studied in the paper.
For each aircraft in the sample we construct two measures of aircraft vintage, which are then
related to the creditor rights scores described below. The first measure is aircraft age, defined in
each year as the time elapsed from the year of the aircraft’s initial delivery. The second measure
of vintage, which we name ‘technological age’, is defined as the time elapsed from the year in
which the aircraft’s model type was first introduced. This second measure proxies for the age of the
technology that is embodied in the aircraft. The Ascend CASE database defines two aircraft-type
classification – narrow and broad. We thus define two variants of technological age corresponding to
the two aircraft type classification. To fix ideas consider the following example: aircraft N368AA,
built in 1991, and delivered on December, 5, 1991 to American Airlines is a Boeing 767-300ER. In
this case, the broad classification is the B767 model type which was first introduced in 1981. This
particular variant of B767 (i.e. 300ER) was first introduced in 1986. Thus, as of the year 2008,
the aircraft’s age is 17 years, its technological age using the broad classification is 27 years, and its
technological age using the narrow definition is 22 years.
Panel A of Table 1 displays summary statistics of aircraft age for 4 sub-periods and for the
entire sample. There are 489,916 aircraft-year observations in the entire sample, with an average
(median) age of 13 (12) years, and a standard deviation of 9.2 years. The sample represents 219
aircraft types, 5,987 operators from 129 countries.
In the last two columns of Panel A we split our sample into aircraft operated by commercial
and private airlines (Commercial), and those operated by the military and government agencies
(Military) – a distinction that plays an important role later in the analysis. There are 373,261
commercial aircraft and 116,655 that are classified as military aircraft in the sample. The commercial sample represents 161 aircraft types, 5,437 operators from 129 countries, while the military
sample represents 200 aircraft types, 893 operators from 115 countries. Further, as can be seen
in Panel A of the table, military aircraft are older than commercial aircraft; the average age of a
commercial aircraft is 12.0 years compared to 16.0 years for military aircraft (p-value for an equal
means t-test=0.000).
8
Benmelech and Bergman (2008a, 2008b) provide an extensive description of the Ascend CASE database.
9
Panel B of Table 1 presents summary statistics for broad and narrow (in parentheses) technological age. The mean broad (narrow) technological age of the entire sample is 21.9 (18.2). As
in Panel A, we split our sample into commercial and military aircraft in the last two columns of
Panel B. Military aircraft embody older technology than commercial aircraft; the average broad
technological age of a commercial aircraft is 20.2 years compared to 27.1 years for military aircraft
(p-value for an equal means t-test=0.000).
B.
Country Level Data
We match the data on individual aircraft to country level macro and legal variables of the aircraft’s
country of operator and owner. We augment the data from Ascend CASE with country level macro
data from the World Bank’s World Development Indicators database. This macro data includes
GDP and GDP per capita in U.S. Dollars, GDP growth, GDP per capital growth, as well as country
area (in sq km.) and population data. We obtain data on legal origins and creditor rights from the
new database assembled by Djankov et al. (2007) that covers 129 countries in the period 1978-2003.
This new data is a major improvement upon the La Porta et al. (1997, 1998) data, as it covers
many more countries and tracks their variation in creditor rights score over time.
For each country, the creditor rights index measures four powers of secured lenders in bankruptcy.9
First, whether there are restrictions on bankruptcy filing; second, whether there is no ‘automatic
stay’ or ‘asset freeze’ that prevent secured creditors from seizing their collateral. Third, whether
secured creditors are paid first, and finally, whether a trustee different from the management runs
the firm during reorganization. A value of one is assigned to each of the provisions when a country’s law provides these powers to secured creditors. The creditor rights index is then calculated by
aggregating the scores of the four provisions, and varies between a score of 0 (poor creditor rights)
and 4 (strong creditor rights). Djankov et. al. (2007) collect time series data on creditor rights
for each of the 129 countries by identifying all major reforms and assessing their impact on the
creditor rights score.
Panel C of Table 1 reports summary statistics of the creditor rights index, GDP per capita,
and legal origin. The mean (median) creditor rights in the sample is 1.64 (1.0) and the standard
deviation is 1.01. The sample includes 276,601 aircraft from countries with English legal origin,
107,587 from countries with French legal origin, 50,982 from Socialist legal origin, 47,070 from
9
See Djankov et. al. (2007) for a comprehensive description of the index and its construction.
10
German legal origin, and 7,676 aircraft from countries with a Nordic legal origin. With a total of
184,122 observations, the U.S. accounts for 37.58% of the sample, followed by the Russian Federation
(37,907 aircraft), U.K. (19,556 aircraft), and Canada (18,406 aircraft). The countries with the least
observations in the data are Bosnia and Herzegovina (44 aircraft), Albania (49 aircraft), Niger (51
aircraft), and Togo (62 aircraft).
C.
Airline Level Data
Finally, we match aircraft information to airline financial data where available. Information on
airline financial data is obtained from Compustat Global. We collect all firms in SIC codes 45004580 and manually match them to the aircraft level data from Ascend CASE. We also supplement
the information with data from Compustat North America for U.S. airlines. After matching Ascend
CASE to Compustat Global and Compustat North America and restricting the sample to the
countries covered by Djankov et al. (2007), we are left with a subsample of 72 airlines from 29
countries, representing a panel of 94,272 aircraft-year observations.
III.
Aircraft Vintage and Usage
We assume that assets of an older vintage are less efficient, either because they are less technologically advanced or due to physical depreciation. We therefore begin our empirical analysis with
motivational evidence testing this assumption in the context of aircraft. Measuring individual aircraft efficiency requires information on inputs (number of seats, men hour, fuel costs, operating
times, routes, etc.) and outputs (number of passengers, revenue, arrival times). We cannot measure
aircraft efficiency directly since we do not have access to these data at the individual aircraft level.
Instead, we utilize data from the Ascend Case database on aircraft usage as an approximation of
aircraft efficiency. Spanning the period 1996-2006, the data provides hourly utilization rates for
25,009 aircraft worldwide. For each aircraft in the sample, the data tallies the number of hours
flown each year, as well as the aircraft type and year of build.
We hypothesize that if aircraft efficiency is indeed decreasing with aircraft vintage, airlines will
tend to decrease the operating times of their older vintage aircraft. Thus, for example, if older
vintage aircraft are less fuel-efficient airlines will shift their operations to the newer vintage aircraft
in their fleet to the extent possible. Moreover, older aircraft require more maintenance and engines
overhauls that would ground older aircraft for longer periods of time compared to newer ones.
11
Figure 3 provides a graphical representation of this monotonic relation between age and usage.
To construct the figures we regress yearly aircraft usage on the set of indicator variables defined for
each possible value of aircraft age, while including year and aircraft-type fixed effects as well. The
figure graphs the coefficients on the age indicator-variables along with their 95 percent confidence
interval calculated by clustering at the aircraft-type level.10 The graph illustrates the evolution of
aircraft usage with aircraft age. Consistent with our assumption that aircraft efficiency improves
over time, aircraft usage declines with aircraft age. We also used aircraft age as an explanatory
variable instead of the set of indicator variables that were used to construct Figure 3. In unreported results we find that the coefficient on aircraft age is consistently negative and is statistically
significant at the 1% level whether we cluster the standard errors by aircraft-type or at the individual aircraft level. Thus, consistent with our underlying assumption that older vintage aircraft
are less efficient, we find that aircraft usage declines with age. This result is robust to the addition
of both aircraft type and year fixed effects. The economic magnitude of this effect is significant:
a one-standard-deviation increase in aircraft age of 8.62 years decreases aircraft yearly usage by
approximately 450 hours, representing an 18 percent decline relative to the sample mean hourly
usage of 2,466 hours.
IV.
A.
Creditors Rights and Aircraft Vintage
Baseline Results
Our simple model shows that the effects of financial constraints should be exacerbated in countries
with poor creditor rights, where the availability of debt capital may be limited and its cost much
higher. We therefore predict that airlines that operate in countries with poorer investor protection
operate older vintage aircraft with older technologies.
To test this prediction, we calculate the age and the technological age (using both the narrow
and broad measures described above) of every aircraft in the 129 countries that are in our sample
during the period 1978-2003. We then run the following specification:
V intageiact = α × Creditor rightsct + Xct λ + ytθ + zac ψ + iact ,
(4)
where the dependent variable, Vintageiact, is either the age or the technological age of aircraft i
operated by operator a in country c in year t. Creditor rights is the creditor rights score of country
10
The indicator variable for age equaling one is omitted, so that all coefficients are calculated in relation to the
usage of aircraft of age one.
12
c in year t, as measured by Djankov, et al. (2007). Xc is a vector of country-specific control
variables which includes the logarithm of country c’s GDP, GDP per capita, and annual rates of
growth of both GDP and GDP per capita, the logarithm of its population and the logarithm of
its area. In addition, in all specifications that do not include country fixed effects we include as
control variables a set of indicator variables indicating the legal origin of the country – common
law, French, German, Nordic, or Socialist.11 Finally, all regressions include year fixed effects, y t,
and depending on the specification may also include country and operator fixed effects represented
by the vector of variables z. Since aircraft operators maintain their affiliation with the country of
operation throughout the sample, country and operator fixed effects are always applied separately
in each specification. All regressions are estimated with heteroscedasticity robust standard errors
which are clustered by country. In our data, standard errors that are clustered by country are
tenfold larger than simple robust standard errors. Thus, when we do not cluster we get a t-statistic
on creditor rights that is between 19.0 and 26.8. Since our variable of interest is creditor rights
which is determined at the country level, we use the higher hurdle of clustering by country. The
magnitudes of the differences between the standard errors when we cluster compared to simple
robust standard errors are consistent with Kloek (1981) who shows that clustered standard errors
√
are proportional to the squared root of the number of clusters which is 129 = 11.4 in our sample.
Table 2 provides results of regression (4) over the entire sample. As hypothesized, we find that
enhanced creditor rights are consistently negatively associated with both aircraft age as well as
aircraft technological age. As the first column of Table 2 demonstrates, with year fixed effects,
increasing a country’s creditor rights score from 0 to 4, reduces the age of aircraft by 1.78 years,
or 13.7% of the mean aircraft age of 13 years. Adding either country or operator fixed effects
(representing the 129 countries in the sample and 5,987 different operators) increases the magnitude
of the negative impact of creditor rights on fleet aircraft age. With these fixed effects, a movement
from a creditor rights score of 0 to a score of 4 reduces aircraft age by between 2.7 and 3.2 years,
representing an approximately 20% reduction in the sample mean aircraft age.
Columns 4 to 6 of Table 2 show that enhanced creditor rights is also negatively related to aircraft
technological age constructed using the broad aircraft classification scheme. This result holds when
using year, country and operator fixed effects. The impact of creditor rights is significant: moving
11
Country fixed effects naturally preclude using legal origin controls as there is no time series variation in legal
origin in our sample period. For brevity of exposition, tables do not exhibit the coefficients on the legal origin dummy
variables.
13
the creditor rights score from 0 to 4 reduces average technological age of aircraft in an airline’s fleet
by between 1.6 and 2.8 years representing between 7.2 and 12.7 percent of the average technological
age.12 Finally, as columns 7-9 show, repeating the analysis using technological age defined at the
narrow classification scheme yields similar results. In sum, consistent with our prediction, aircraft
are younger and embody newer technology in countries with better creditor rights, controlling for
GDP per capita, population, area, and a battery of fixed-effects at the operator, country and year
level. We repeat the analysis in Table 2 by calculating average aircraft age within a country for each
of the years, thereby collapsing the data to the country level, and estimating weighted least-squares
regressions. These regressions which are not reported for brevity yield similar results.
The negative relation between creditor rights and both aircraft age and aircraft technological
age points to a financing channel through which improved investor protection and its associated
reduction in financial frictions affects firm investment policy and ultimately real outcomes. According to this, the ability to raise external finance is an important determinant of firm’s capacity
to invest in newer technologies which is a key driver of economic growth.
B.
Identification Strategy
As is usually the case in cross-country analysis, the main empirical challenge is endogeneity and,
in particular, an omitted variable bias problem. Specifically, the creditor rights score could be
correlated with other unidentified variables, such as investment opportunities, which are in turn
influencing asset vintage choices. The relation between creditor right and vintage could then be
explained by effects other than the financing channel proposed in this paper. Most of the analysis
that follows is aimed, therefore, at addressing the direction of causality in the empirical findings of
Table 2.
To overcome the omitted variables problem we first utilize the panel nature of our data and the
changes in creditor rights over time. By including country and operator fixed effects, we control for
unobserved and non-time varying heterogeneity of operators and countries. In these specifications
we identify off of changes in creditor rights over time within a country. Indeed, we find that in
specifications which include country fixed effects, the negative association between creditor rights
and age is the largest, which is consistent with a large effect of changes in creditor rights within a
12
As in columns 1 to 3, the magnitude of the effect is larger when either operator or country fixed effects are
included.
14
country.
However, while country fixed-effects help to mitigate concerns about unobserved heterogeneity, including these fixed-effects raises the issue of the endogeneity of creditor protection laws
themselves. For example, a country may revise its corporate and bankruptcy laws precisely when
underlying economic conditions improve. In this case the correlation between creditor rights and
aircraft vintage may merely reflect increased demand for better aircraft driven by improvements in
economic conditions.
One solution for the endogeneity concern is to use an instrumental variable approach. However,
variables that are correlated with creditor rights are also potentially correlated with aircraft age
through channels other than the law, and hence will not meet the ‘exclusion’ restriction. Consider,
for example, legal origins (as in La Porta et. al 1998) as an instrument for creditor rights. While
correlated with creditor rights, legal origins are potentially correlated with aircraft vintage through
other legal and economic mechanisms such as safety regulation or engineering quality. Djanakov
et. al (2007) raise similar concerns about the validity of legal origin as an instrument for creditor
rights in general.
In the absence of an instrument, we identify the causal effect of creditor rights on aircraft age
by splitting our sample into aircraft that are expected to be treated by stronger creditor rights and
those that should not. Our identification strategy is threefold. First, we split the sample between
commercial and military aircraft and show that, as expected, military aircraft are not treated by
creditor rights. Second, we focus on commercial aircraft and split the sample between leased and
owned aircraft. Since prior literature has shown that leasing allows firms to relax financial frictions
(see e.g. Eisfeldt and Rampini, 2008), the financing channel predicts that the negative relation
between creditor rights and vintage should be concentrated amongst non-leased aircraft. Our
empirical results confirm the prediction that leased aircraft are not treated by the creditor rights
index similarly to non-leased aircraft. Finally, we study the differential effect of creditor rights
on commercial aircraft conditional on the financial health of the airline. The financing channel
predicts, and we indeed find, that the relation between creditor rights and vintage is concentrated
amongst airlines that are in poorer financial health.
15
C.
Commercial vs. Military Aircrafts
In Table 3, for every country, we divide our sample into aircraft operated by commercial airlines
and private operators, and those operated by militaries, armed forces and government agencies.
For example, as of December 31, 2003 there are ten U.S. federal agencies or military operators in
our sample: Federal Aviation Administration, NASA, US Air Force, US Air National Guard, US
Army, US Army National Guard, US Coast Guard, US Customs Service, US Marine Corps, and
the US Navy.
We expect the negative relation between aircraft age and creditor protection to be concentrated
in commercial and private aircraft operators, since these are the firms which would be required
to raise funds from external investors in cases of financial shortfalls. In addition, commercial
and private firms would fall under the bankruptcy provisions of the local corporate bankruptcy
laws which are the essence of the creditor protection score. In contrast, government agencies,
militaries and other armed forces obtain funding from their governments that are in turn subject
to international law. However, when sovereign governments default on their debt creditors cannot
effectively seize the country’s assets.13 Instead, creditors litigate with their sovereign borrowers
using international law. Sovereign borrowers are then induced to settle with their creditors as they
want to maintain access to capital markets (Bulow and Rogoff (1989a,b)). In summary, corporate
and bankruptcy law do not apply to aircraft operated by militaries, armed forces and government
agencies.
To formally test the hypothesis that the negative relation between aircraft age and creditor
protection will be concentrated in commercial and private aircraft operators, Table 3 reports the
results of running regression 4 separately for commercial aircraft and aircraft operated by militaries
and other government agencies.14 The dependent variable is aircraft age and the explanatory
variables are as in Table 2 with the addition of the country’s airforce size.
Consistent with the results in Table 2, in all specifications, the age of commercial aircraft is
statistically negatively related to creditor rights with a similar economic impact found in Table 2,
while the age of military aircraft is unrelated to the creditor rights score in a statistically significant
13
Although lenders can potentially grab commercial or military and government-owned aircraft, this strategy is
not very useful except as a strategy of harassment (Shleifer 2003).
14
This specification is identical to running one regressions with all the explanatory variables interacted with a
military dummy. We prefer to report the results separately for commercial and military aircraft as the exposition
is clearer. Table 10 reports tests on the statistical significance of the difference between commercial and military
aircraft.
16
manner. Table 4 repeats the analysis, separating the sample into commercial aircraft and aircraft
operated by militaries and other government agencies, but this time using aircraft technological
age as the dependant variable. Again, we find that while aircraft technological age, defined using
either the narrow or broad classification, is negatively related to creditor rights in the subsample
of commercial aircraft, there is no statistically significant relation between the technological age of
military aircraft and country creditor rights scores. Interestingly, while not statistically significant,
we find that in the subsample of military aircraft, the coefficients on creditor right scores are
actually positive.
D.
Creditor Rights and Aircraft Leasing
In the analysis above, we do not distinguish between airlines that lease aircraft instead of purchasing
them through debt financing.15 Eisfeldt and Rampini (2008) show that, in the U.S., since the
repossession of leased assets is easier than foreclosure on collateral of secured debt, lease financing
allows for higher debt capacity than secured debt. Put differently, lease financing aids firms to
circumvent some of the financial frictions associated with debt financing. To the extent that this
result generalizes to other countries outside the U.S. – for example, because the title to the asset
remains with the lessor but is not in the possession of a secured creditor – we would expect two
implications to arise. First, airlines operating in countries with poor creditor rights should be
more likely to use lease financing rather than plain debt because of the associated reduction in
financial frictions. Second, if leasing reduces financial frictions, then the results found in Tables
2 - 4 showing that creditor rights is negatively related to aircraft vintage should be stronger for
non-leased aircraft.
To test the first conjecture, we run a probit regression on the sample of all commercial aircraft,
relating a country’s creditor rights score to the likelihood of an individual aircraft being leased:
P r(leased = 1) = Φ(−0.026 ∗ Creditor Rightsct + Xct λ + yt θ),
(5)
where Φ(·) is the standard normal cumulative distribution function, Creditor rights is the creditor
rights score of country c in year t as measured by Djankov et al. (2007), Xc is the vector of control
variables used in regression 4, and yt is a vector of year fixed effects. The point estimate of -0.026
(p-value=0.04) implies that airlines in countries with poor creditor rights are indeed more likely
15
Lease financing of aircraft is fairly common, particularly in the United States (See e.g. Benmelech and Bergman
2008a and Gavazza 2007).
17
to lease their aircraft. This effect is economically significant: moving from a creditor rights score
of 4 to a creditor rights score of 0 increases the likelihood that an aircraft will be leased by 10.4
percentage points, representing an increase of 24.2 percent relative to the unconditional mean.
Thus, the data do indeed suggest that airlines operating in countries with low creditor protection
are more likely to resort to lease financing.
To test the second conjecture, that the negative relation between creditor rights and aircraft
vintage should be stronger for non-leased aircraft, we repeat the analysis in regression 4 separately
for leased and non-leased aircraft.16 Results are reported in Table 5.
Consistent with our conjecture, we find that the negative relation between aircraft vintage and
creditor rights is indeed concentrated amongst owned, rather than leased, aircraft. For owned
aircraft, a zero to 4 increase in creditor rights associated with a reduction of between 2.6 and 3.9
years in aircraft vintage. In contrast, in the leased aircraft subsample, in all of the specifications
the coefficient on creditor rights is not statistically different from zero.
Another benefit to examining the effect of creditor right on leased and non-leased aircraft
separately is that it alleviates the concern that our results are driven by variation in omitted
variables and in particular variation in investment opportunities correlated with variation in creditor
rights. While this concern is partly addressed by our battery of GDP-based controls, and operator
and country fixed effects specifications, to the extent that time series variation in creditor rights is
correlated with investment opportunities – for example because bankruptcy reform may be enacted
simultaneously with other economic reforms (see e.g. Acharya and Subramanian, 2008) – we
cannot completely rule out the possibility that variation in investment opportunities is driving the
results. However, the fact that the negative relation between creditor rights on aircraft vintage is
concentrated in non-leased aircraft alleviates this concern. There is little reason to suspect that
increased investment opportunities should differentially impact the vintage of leased as compared to
non-leased aircraft, while the financing channel provides a clear prediction regarding the differential
impact of creditor rights on the two methods of aircraft financing.
16
As before, this specification is isomorphic to running one regressions with all the explanatory variables interacted
with a leasing dummy. Table 8 reports test on the statistical significance of the difference between leased and
non-leased aircraft.
18
E.
Creditor Rights and Aircraft Leasing: Poor vs. Rich Countries
Table 6 revisits the results of the previous table for richer (higher or equal to GDP per capita
median) and poorer (below GDP per capita median) countries separately. This allows us to control
better for economic development, by not only including all four GDP-based measures of development, but also allowing for a more flexible functional form in the relation between aircraft age and
economic development. As can be seen in the table, we find that among the development variables,
GDP per capita is the strongest predictor of aircraft age across most of the specifications with
higher GDP per capita growth associated with younger aircraft.
More importantly, we find that the negative relation between creditor rights and aircraft age
is driven by rich countries (the coefficient on creditor rights is between -0.648 and -1.017), while
for poor countries the coefficient is between -0.025 and 0.345 and is not statistically different from
zero. There are fewer aircraft in the poor countries sample compared to the rich countries sample
(35,313 vs. 177,937 observations). However, since we cluster the standard errors by country and
given that there are roughly the same number of countries in each of the sub-samples, the results
are not driven by lack of statistical power in the poor countries regressions. Instead, our results
are consistent with Djankov et al. (2007) who find that creditor rights have an impact on credit
markets only in richer countries. In contrast, creditor rights in poorer countries have little impact
on credit market development, possibly due to their lack of enforcement.
Finally, in the last three columns of Table 6 we focus on non-leased aircraft of rich countries
excluding all U.S. aircraft, which account for 37.58% of all aircraft in our sample. We exclude U.S.
aircraft as a robustness check to verify that our results are not driven by specific characteristics of
U.S. airlines that are potentially correlated with creditor rights. While the result in the specification
with only year fixed-effects is slightly less significant compared to the first column of the table, the
coefficients are similar whether the U.S. is included or excluded. Taken together, our results thus
suggest that creditor rights are more important in richer countries and are not an artifact of the
U.S. legal system.
F.
Robustness Tests: Government Ownership and Regulation
We now turn to check the robustness of our results in several ways. First since many commercial
airlines are either fully or partially state-owned we need to control directly for government ownership
in our regressions. One concern about government ownership is that it is correlated with creditor
19
rights and hence potentially biasing our point estimates. Similarly, differences in aviation regulation
may influence aircraft vintage while at the same time be correlated with a country’s creditor rights
index. Thus, we need direct measures of both government ownership at the airline level, and
aviation regulation quality at the country level.
F.1
Government Ownership, Creditor Rights and Aircraft Vintage
The financing channel would predict that commercial airlines with government ownership can utilize
the government as a source of capital to ease financing constraints. Commercial airlines wholly or
partially owned by the government may have a ‘soft’ budget constraint and as a result should have
fleets of younger vintage. Furthermore, governments may be willing to invest more in new aircraft
in their ‘flag carriers’ as they represent the country internationally. According to Littlejohns and
McGairl (1998): “Because a new aircraft symbolizes not only the nation’s prestige, but also the
skill of its leaders, it is easy for politicians to value these symbols far above mere prudence.”17
We collect data on government ownership in airlines from Ascend Case, and supplement it with
information from airline web sites and Lexis-Nexis to construct a dummy variable taking on the
value of 1 for airlines with some government ownership in a particular year, and zero otherwise.
We then run regressions similar to the specification in regression 4 for all commercial aircraft
with the government ownership dummy as an additional control. The sample is divided into
leased and non-leased aircraft, and results are presented in Table 7. First, as hypothesized, and
consistent with a financing channel for vintage capital, government ownership is negatively related
to aircraft vintage, suggesting that governments do indeed relieve some of the financial constraints
of the airlines which they own. Moreover, even after controlling for government ownership, creditor
rights is negatively related to both aircraft age and technological age of non-leased aircraft. The
coefficients of creditor rights in the different specifications (between -0.619 and -1.067) are generally
higher than those found in the panel data regressions.
F.2
Aviation Regulation, Creditor Rights and Aircraft Vintage
As an additional robustness test, in Table 8, we add to the regressions in Table 7 a measure of the
quality of aviation regulation for country c at year t. We construct this measure using information
from the Federal Aviation Administration (FAA) that ranks the overall quality of a country’s civil
17
Littlejohns and McGairl (1998) p. 216.
20
aviation authority. FAA inspectors assess civil aviation authorities around the world based on
their authorities to license and oversee air carriers in accordance with International Civil Aviation
Organization (ICAO) aviation safety standards. The FAA classifies countries into two: those that
are compliant with ICAO standards and those that are not compliant with ICAO standards.18
We were able to obtain data on FAA classifications for 67 countries for the years 1994-2003,
resulting in a total of 188,142 aircraft. We construct a dummy variable measuring aviation quality
which takes on the value of 1 for countries that comply with ICAO standards in a particular year,
and zero for those that fail to comply. We then run similar specifications to regression 4 adding
the aviation quality dummy as an explanatory variable.19 Column 1 reports results using the
sample of all the commercial aircraft, in columns 2 and 3 the sample is divided into leased and
non-leased aircraft, and in column 4 we use only military aircraft. The first four columns of Table
8 confirm our previous findings. After controlling for both government ownership and aviation
regulation quality, it is only the vintage of non-leased commercial aircraft which is related to the
creditor rights index while the vintages of leased commercial aircraft and military aircraft are not.
Furthermore, controlling for government ownership and aviation regulation increases the impact of
creditor rights as compared to our previous estimates – moving from a creditor rights score of 0 to
a creditor rights score of 4 reduces non-leased commercial aircraft age by 7.38 years.
In the last 2 columns of Table 8 we test the significance of the difference between the coefficient
on creditor rights in: (i) the military and commercial non-leased aircraft subsamples; and (ii)
the commercial leased and non-leased aircraft subsamples. To do so, we first add an interaction
coefficient between the creditor rights score and a dummy variable capturing whether an aircraft
is a military aircraft to the previous regressions (Columns 1-4). As can be seen, the interaction
coefficient is positive and significant at the 10% level, indicating that the negative relation between
creditor rights and the vintage of non-leased aircraft is statistically different than that between
creditor rights and the vintage of military aircraft. In Column 6 we add an interaction term
between creditor rights and a dummy variable taking on the value of one if an aircraft is leased.
Again, the positive coefficient on the interaction term demonstrates that the difference between the
18
According to the FAA a country fails to comply if one the following deficiencies are identified: (i) lack of laws
or regulations necessary to support the certification and oversight of air carriers, (ii) lack of the technical expertise,
resources, and organization to license or oversee air carrier operations; (iii) lack of adequately trained and qualified
technical personnel; (iv) lack of enforcement or compliance with, minimum international standards; and (v) insufficient
documentation and records of certification of air carrier operations.
19
We do not include operator or country fixed-effects in these regressions since the aviation quality measure hardly
changes over time within a country.
21
coefficients on the creditor rights variable in the leased and non-leased subsamples (Columns 2 and
3) is statistically significant.
G.
Creditor Rights, Financial Constraints and Aircraft Vintage
We now turn to analyze the effect of creditor rights on aircraft age conditional on the financial
condition of the operator. Since airlines with greater internal funds are less likely to rely on
external financing, they should be less affected by the legal system in which they operate or local
financial development. Thus, we expect the effect of creditor rights on aircraft age to be larger for
more financially constrained airlines.
Similar to the previous section, testing this prediction alleviates the concern that creditor rights
are positively correlated with unobserved investment opportunities, and that it this correlation
which is driving the negative relation between creditor rights and aircraft vintage. This is because there is little reason to suspect that increases in creditor rights are more strongly correlated
with improved investment opportunities in financially constrained firms as compared to financially
unconstrained firms.
To test this prediction, we obtain information on airline financial data from Compustat Global.
We are able to match 67 airlines from 28 countries to the countries covered by Djankov et al. (2007),
representing a panel of 63,036 non-military aircraft. We then employ in our regression specification
interaction terms between the country’s creditor rights index and airline-level measures of financial
distress. Our approach is similar to Rajan and Zingales (1998) who identify the effects of financial
development on growth using interaction terms between financial development (at the country level)
and financial dependence (at the industry level). Our analysis focuses on two measures of financial
constraints: leverage and long-term debt, both used by Eisfeldt and Rampimi (2007) which were
found empirically to be determinants of used capital investment. We obtain similar results using
other measures such as profitability. We estimate the following regression for both leased and
non-leased aircraft separately:
V intageiact = α × Creditor rightsct + β × F inConstact
(6)
+ γ × Creditor rightsct × F inConstact + Xct λ + yt θ + zac ψ + iact ,
Where F inConstact is a measure of the airline financial constraints (either leverage defined as total
debt divided by the book value of assets, or long-term debt defined as long-term debt divided
22
by the book value of assets), and Creditor rightsct × F inConstact is an interaction term between
creditor rights and airline financial constraints. Regression are estimated with heteroscedasticity
robust standard errors clustered by country. Results using aircraft age as the dependent variable
are presented in Table 9.20
Consistent with the financing channel, the interaction term between creditor rights and leverage and the interaction term between creditor rights and long-term debt in the non-leased aircraft
subsample (Columns 1, 3, 5, 7) are negative, indicating that the effect of creditor rights on aircraft age is indeed concentrated in financially constrained airlines. In contrast, consistent with our
previous results and consistent with a financing channel, we do not find a statistically significant
interaction coefficient between creditor rights and measures of financial constraints in the leased
aircraft subsample. While both leverage and long-term debt are clearly endogenous, our identification strategy in Table 9 relies on the interaction between country and firm characteristics. By
focusing on interaction effects we reduce the number of potential alternative explanations for our
findings.
Focusing on the first column of Table 9, we find that reducing a country’s level of investor
protection from a creditor-rights score of 4 to a creditor rights score of 0, increases the average age
of aircraft operated by airlines in the 25th percentile of leverage by 0.95 years. In contrast, for
airlines in the 75th leverage percentile, i.e. those that are arguably more financially constrained,
we find that reducing creditor rights from a score of 4 to a score of 0 increases average age by
four times as much – 3.98 years, representing a 30 percent increase compared to the sample wide
average of aircraft age.
H.
Creditor Rights and Fleet Size
We now analyze the relation between creditor rights and fleet size. According to Prediction 2
of the model, firms operating in countries with better creditor rights should operate larger fleets
on average. This is because operator scale will not be constrained by the availability and cost
of external finance. This prediction is broadly consistent with the empirical findings in Kumar,
Rajan, and Zingales (2002) who find that the average firm size is larger in countries with better
institutional development.
20
We obtain similar results using aircraft technological age as the dependent variable. We do not report these
results for brevity.
23
In order to test this prediction we need a measure of fleet size. This is somewhat complicated
by the fact that airline fleets include multiple aircraft types of different size and use. Thus, a
measure of fleet size must weigh aircraft of different variety in an appropriate manner. Rather
than committing to one particular weight system, we test Prediction 2 using a number of weighing
schemes. To do so, for each aircraft type in our sample, we gather information on that aircraft
type’s maximal seat capacity, its maximal takeoff weight, and the aircraft type’s wingspan. This
data is gathered from Singfield (2005) as well as from a variety of Internet sources. Based on this
information, for each operator and year in our sample, we then construct four measures of fleet size.
The first is simply an equal-weighted sum of all aircraft operated. The remaining three measures
of fleet size are: (1) the sum of the seat capacities of all aircraft in the fleet, (2) the sum of the
maximal takeoff weight of all aircraft in the fleet, and (3) the sum of the wingspans of all aircraft
in the fleet.
Having constructed these four fleet-size measures, we then run the following regression specification for all operator fleets in our sample period of 1978-2003:
log(Sizeact) = α × Creditor rightsct + Xct λ + yt θ + za ψ + iact ,
(7)
The dependant variable, log(Sizeact), is the logarithm of each of our four fleet-size measures for
operator a in country c in year t. As usual, Creditor rights is the creditor rights score of country
c in year t, and Xc is the standard vector of country-specific control variables. All regressions
include year fixed effects, y t, and operator fixed effects represented by the vector of variables za .
The regressions are estimated with heteroscedasticity robust standard errors clustered by country.
Finally, as in the case of aircraft age, regression 7 is estimated separately for commercial operators
and military operators.
The results are provided in Table 10. As our results demonstrate, using all four fleet-size proxies,
the coefficient on the creditor rights index is consistently positive and statistically significant in
the commercial operators regressions after controlling for GDP per capita, population, area, as
well as year and operator fixed effects. In contrast to commercial operators, and consistent with
our previous results, there is no robust relation between creditor rights and fleet size of military
operators. None of the creditor rights coefficients are statistically different from zero in any of the
military operators regressions, and the point estimates in these regressions are always lower than
their commercial regressions counterparts (they are actually negative in 3 out of 4 regressions).
24
Thus, consistent with prediction 2 of the model, airlines in countries with higher creditor rights
do indeed operate larger fleets. Given that we run a semi-log specification with respect to creditor
rights, the coefficient of creditor rights is equal to the percentage change in fleet size associated with
a unit change in creditor rights (dlog(Size)act/dCRct ). This effect is economically significant. For
example, moving from the lowest creditor rights score of zero, to the highest score of four, increases
the number of aircraft in a commercial airline’s fleet by 22 percent. Moving from the lowest to the
highest score of creditor rights increases total fleet seat capacity by 72.8 percent. The effect for the
remaining two fleet size measures – total fleet maximal takeoff weight and total fleet wingspan – is
73.2 and 55.2 percent, respectively.
Finally, we study the relation between fleet size and creditor rights conditional on the financial
constraints that an airline is facing. In results not reported, we find no evidence that the positive relation between creditor rights and fleet size exhibited in Table 10 is concentrated amongst
financially constrained airlines. This could stem from a selection bias arising from the fact that
accounting data availability constrains our analysis to publicly traded airlines, which tend to be the
largest airlines in each country. Alternatively, this could be driven by the positive relation which
arises between leverage and fleet size when airlines use debt to purchase additional aircraft.
V.
Conclusion
We provide novel evidence linking creditor rights and vintage capital using a panel of aircraft-level
data around the world. Consistent with theories that emphasize the protection of property rights as
essential for economic development, we find that better creditor rights are associated with aircraft
of a younger vintage and firms with larger aircraft fleets. Further, consistent with a financing
channel, we find that the association between creditor rights and aircraft vintage is concentrated
amongst non-leased commercial aircraft. Finally, we find that airlines with lower leverage ratios
and airlines with less debt overhang, are less sensitive to creditor rights as they may use internal
funds, rather than external capital, to finance investment.
The evidence in our paper shows that legal protection of creditor rights affects both capital
vintage, technological diffusion and firm scale. Better creditor protection helps airlines to mitigate financial short-falls and enhance investment in newer, more efficient and more technologically
advanced aircraft. While we study the relation between vintage aircraft and creditor rights, our re-
25
sults propose a broader link, not confined only to the airline industry, between investor protection,
real corporate investment and economic growth; legal protection of creditors facilitates the ability
of firms to make large capital investments, adapt advanced technologies and fosters productivity.
Appendix
Proof of Proposition 1. Define z(qnew ) = pold ∗ h(qnew ) to be the price of obtaining revenue
f (qnew ) using a fleet of old technology aircraft. For simplicity in what follows we drop the subscript
and denote qnew by q. If z (0) ≥ 1, from the convexity of z we have that z(q) > q for all q > 0 and
the proposition trivially holds with μ̄ = 0. Similarly, if z (0) < 1 and z(q) < q for all q > 0 the
proposition trivially holds with μ̄ = 1. Assume then that z (0) < 1 and that there exists a q > 0
with z(q) = q. By the convexity of z, this q is unique, and we denote it by q ∗ . Clearly z(q) < q for
q < q ∗ and z(q) > q for q > q ∗ .
Now, for any μ we define the ‘financable set’ to be all q with c(q) ≤ μf (q), where c(q) = min(z(q), q).
Since f is concave, z is convex, f (0) = 0, and z(0) = 0 it is easy to see that for any μ the financable
set equals [0, q] for some q. Now, define μ∗ to satisfy μf (q) = q ∗ . By definition of q ∗ , the financable
set at μ∗ is [0, q ∗]. Since in this region z(q) < q for all q which are financable, a firm operating in
a country with creditor protection μ∗ chooses the old technology fleet. Similarly, for any μ < μ∗ ,
the financable set equals [0, q ] for some q < q ∗ , so that again, z(q) < q for all financable q. Thus,
again, any firm operating in a country with creditor protection μ < μ∗ chooses the old technology.
Define q uc to be the solution to the unconstrained problem M axq [f (q) − c(q)]. Since by assumption
the new technology is preferred to the old when μ = 1, we have that c(q uc ) = q uc , so that q uc > q ∗ .
Define μ∗∗ to satisfy μf (q) = q uc . Clearly, for any μ ≥ μ∗∗ , q uc is financable, so that a firm operating in a country with μ > μ∗∗ chooses the unconstrained solution and hence the new technology.
Now, for all μ∗ < μ < μ∗∗ the financable set equals [0, q̄(μ)] for some q∗ < q̄(μ) < q uc . Define
V (μ) to be the solution to the maximization problem of a firm operating in a country with creditor
protection μ∗ < μ < μ∗∗ . That is, V = M axq [f (q) − c(q)] s.t. q ≤ q̄(μ). Further, define V1 as
the solution to the maximization problem M axq [f (q) − c(q)] s.t. q ≤ q ∗ , and V2 (μ) as the solution
to the maximization problem M axq [f (q) − c(q)] s.t. q ∗ ≤ q < q̄(μ). Clearly, V = max[V1 , V2(μ)]
and a firm in country μ will choose the new technology iff V2 (μ) > V1 . Now, it is easy to see that
V2 (μ) is increasing in μ. Also, since the new technology is preferred to the old with no constraints,
we have that V2 (μ∗∗ ) > V1 . Thus, since V1 is independent of μ, there exists a μ∗ ≤ μ̄ < μ∗∗ such
that for all μ∗ ≤ μ ≤ μ̄, V1 ≥ V2 (μ) and for all μ̄ ≤ μ ≤ μ∗∗ , V2 (μ) ≥ V1 . Thus, firms in countries
with μ∗ ≤ μ ≤ μ̄ choose the old technology, while firms in countries with μ̄ < μ ≤ μ∗∗ choose the
new technology. This, combined with the fact that firms in countries with μ ≤ μ∗ choose the old
technology, while firms in countries with μ ≥ μ∗∗ choose the new technology proves the proposition.
Proof of Propositions 2 and 3. Proposition 2 is a direct consequence of Proposition 1, and
Proposition 3 is a direct consequence of the fact that as μ tends to zero, the fleet size that is
financable under the constrained maximization problem tends to zero as well.
26
References
Acharya, Viral V., and Krishnamurthy Subramanian, “Bankruptcy Codes and Innovation,” Review
of Financial Studies, Forthcoming (2008).
Acharya, Viral V., Yakov Amihud, and Lubomir Litov, “Creditor Rights and Corporate Risktaking,” Working Paper (2008).
Beck, T., R. Levine, and N. Loayza, “Finance and the Sources of Growth,” Journal of Financial
Economics, 2000.
Benhabib, Jess and Aldo Rustichini, “Vintage Capital, Investment, and Growth,” Journal of Economic Theory, 55 (1991), 323-339.
Benmelech, Efraim, and Nittai K., Bergman, “Liquidation Values and the Credibility of Financial
Contract Renegotiation: Evidence from U.S. Airlines ,” Quarterly Journal of Economics, Forthcoming (2008a).
Benmelech, Efraim, and Nittai K. Bergman “Collateral Pricing,” Journal of Financial Economics,
Forthcoming (2008b).
Bergman, Nittai, K., and Daniel Nicolaievsky “Investor Protection and the Coasian View,” Journal
of Financial Economics, 2007.
Braun, Matias, “Financial Contractibility and Assets’ Hardness,” Mimeo, 2003.
Boucekkine, Raouf, David de la Croix, and Omar Licandro, “Vintage Capital,” New Palgrave
Dictionary in Economics, Second Edition, (2008) Forthcoming.
Bulow, Jeremy, and Kenneth Rogoff “A Constant Recontracting Model of Sovereign Debt,” Journal
of Political Economy, 1989a, 97(1) 155-178.
Bulow, Jeremy, and Kenneth Rogoff “Sovereign Debt: Is to Forgive to Forget?” American Economic
Review, 1989b, 79(1) 43-50.
Chari, V. V., and H. Hopenhayn, “Vintage Human Capital, Growth and the Diffusion of New
technology,” Journal of Political Economy, 99 (1991), 1142-1165.
Demirguc-Kunt, Asli, and Vojislav Maksimovic, “Law, Finance, and Firm Growth,” Journal of
Finance, 53(6) (1998), 2107-2137.
Djankov, S., C. McLiesh, and Andrei Shleifer, “Private Credit in 129 Countries,” Journal of Financial Economics, (2007).
Eisfeldt, Andrea, and Adriano Rampini, “New or Used? Investment with Credit Constraints,”
Journal of Monetary Economics, 54 (2007) 2656-2681.
Eisfeldt, Andrea, and Adriano Rampini, “Leasing, Ability to Repossess, and Debt Capacity,” Review of Financial Studies, Forthcoming (2008).
Esty, Benjamin, and William Megginson, “Creditor Rights, Enforcement, and Debt Ownership
Structure: Evidence from the Global Syndicated Market,” Journal of Financial and Quantitative
Analysis, 38(1) (2003), 37-59.
Fisman, Raymond, and Inessa Love, “Financial Development and Intersectoral Allocation: A New
Approach,” Journal of Finance, 59(6) (2004), 2785-2807.
Gavazza, Alessandro, “Asset Liquidity and Financial Contracts: Evidence from Aircraft Leases,”
Working paper, Yale University (2007).
27
Greenwood J., Z., Hercowitz, and P. Krusell, “Long-Run Implications of Investment Specific Technological Change,” American Economic Review, 87 (1997), 342-362.
Guiso, Luigi, Paola Sapienza, and Luigi Zingales, “Does Local Financial Development Matter?,”
Quarterly journal of Economics, 119(3) (2004), 929-969.
Hsieh, Chang-Tai, “Endogenous Growth and Obsolescence,” Journal of Development Economics,
66 (2001), 153-171.
Jovanovic, Boyan, “Vintage Capital and Inequality,” Review of Economic Dynamics, 1 (1998) 497530.
King, Robert, and Ross Levine, “Finance and Growth: Schumpeter Might Be Right,” Quarterly
Journal of Economics 108 (1993), 717-738.
Kloek, T., “OLS Estimation in a Model where a Microvariable is Explained by Aggregates and
Contemporaneous Disturbances are Equicorrelated,” Econometrica 49 (1981), 205-207.
Kumar, Krishna B., Raghuram G., Rajan, and Luigi Zingales, “What Determine Firm Size?,”
Working Paper University of Chicago GSB, (2002).
La Porta, R., F. Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny, “Legal Determinants of
External Finance,” Journal of Finance, (1997).
La Porta, R., F. Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny, “Law and Finance,” Journal
of Political Economy, (1998).
La Porta, R., F. Lopez-de-Silanes, and Andrei Shleifer, “The Economic Consequences of Legal
Origin,” Working Paper Harvard University, (2007).
Lerner, Josh, and Antoinette Schoar, “Does Legal Enforcement Affect Financial Transactions? The
Contractual Channel in Private Equity,” Quarterly Journal of Economics, 120(1) (2005), 223-246.
Liberty, Jose, and Atif Mian, “Collateral Spread and Financial Development,” Working paper,
University of Chicago (2007).
Littlejohns, Andrew, and Stephen McGairl, eds., Aircraft Financing (third edition), (Euromoney
Books, England), 1998.
Ongena, Steven, and David Smith, “What Determines the Number of Bank Relationships? CrossCountry Evidence,” Journal of Financial Intermediation, 9(1) (2000), 26-56.
Qian, Jun, and Philip E., Strahan. “How Law and Institutions Shape Financial Contracts: The
Case of Bank Loans,” Journal of Finance, (2007) Forthcoming.
Rajan, Raghuram G., and Luigi Zingales, “Financial Development and Growth,” American Economic Review, 88 (1998), 559-586.
Solow, R., “Investment and Technological Progress,” In K. Arrow, S. Karlin, and P. Suppes (eds.)
Mathematical Methods in Social Sciences 1959, (1960), 89-104. Stanford University Press.
Shleifer, Andrei, “Will the Sovereign Market Survive?,” American Economic Review, 93(2) (2003),
85-90.
Shleifer, Andrei, and Robert W. Vishny, “Liquidation Values and Debt Capacity: A Market Equilibrium Approach,” Journal of Finance, 47 (1992), 143-66.
Singfield, Tom, ”Airliners Worldwide – 2nd Edition,” Midland Publishing, 2005
Wurgler, Jeffrey, “Financial Markets and the Allocation of Capital,” Journal of Financial Economics, 58 (2000), 187-214.
28
f (qnew)
qnew
qnew
q*new
Figure 1:
Combined cost and revenue functions of new and old technologies.
V New(μ)
Value
V Old(μ, pOld)
Old Tech
Figure 2:
1.0
New Tech
New and old technology value functions.
29
μ
0
1
4
7 10 13 16 19 22 25 28 31 34 37 40 43 46
-500
Hours
-1000
-1500
-2000
-2500
-3000
Age (years)
Figure 3:
Annual hourly utilization as a function of aircraft age. Regression coefficients are calculated using year and
aircraft-type fixed effects. 95% confidence intervals are calculated using standard errors that are clustered by aircraft type.
30
Table 1:
Summary Statistics
1978-1979
Aircraft age
Minimum
25th Percentile
Mean
Median
75th Percentile
Maximum
Standard deviation
# of Aircraft types
# of Operators
# of Countries
# of Observations
0
4
9.2
10
13
32
5.5
96
969
89
18,854
Panel A: Aircraft Age
1980-1989
1990-1999
2000-2003
0
5
11.0
11
17
42
7.4
136
2,134
102
128,455
0
5
13.5
12
21
52
9.4
196
4,051
129
235,860
0
6
14.7
13
22
56
10.6
202
3,133
129
106,747
Panel B: Technological Age
1978-1979
1980-1989
1990-1999
2000-2003
Broad technological age (Narrow technological age)
Minimum
0 (0)
0 (0)
0 (0)
0 (0)
25th Percentile
13 (9)
15 (10)
14 (11)
16 (13)
Mean
16.1 (12.6)
19.3 (15.3) 22.7 (19.0)
24.2 (20.7)
Median
16 (12)
20 (16)
24 (19)
22 (19)
75th Percentile
20 (17)
24 (20)
30 (27)
34 (30)
Maximum
32 (32)
43 (43)
52 (52)
56 (56)
Standard deviation
5.2 (5.1)
7.3 (6.9)
9.9 (9.5))
11.6 (11.1)
# of Aircraft types
96
136
196
202
# of Operators
969
2,134
4,051
3,133
# of Countries
89
102
129
129
# of Observations
18,854
128,455
235,860
106,747
Panel C: Country Characteristics
GDP
English
French
Creditor rights per capita legal origin legal origin
Minimum
0
$82.16
0
0
25th Percentile
1
$2,122.3
0
0
Mean
1.64
$17,037.1
0.57
0.22
Median
1
$19,591.6
1
0
75th Percentile
2
$28,262.6
1
0
Maximum
4
$45,390.5
1
1
Standard deviation
1.01
$12,410.4
0.50
0.41
Number of Observations by legal origin
276,601
107,587
Full Sample
Commercial
Military
0
6
13.0
12
19
56
9.2
219
5,987
129
489,916
0
5
12.0
11
18
56
8.8
161
5,437
129
373,261
0
8
16.0
15
23
47
9.8
200
893
115
116,655
Full Sample
Commercial
Military
0 (0)
14 (11)
21.9 (18.2)
21 (17)
29 (25)
56 (56)
9.8 (9.4)
219
5,987
129
489,916
0 (0)
13 (10)
20.2 (16.7)
20 (16)
27 (23)
56 (56)
9.1 (8.8)
161
5,437
129
373,261
0 (0)
20 (15)
27.1 (22.7)
27 (22)
35 (30)
47 (47)
10.1 (10.0)
200
893
115
116,655
German
legal origin
0
0
0.10
0
0
1
0.30
47,070
Nordic
legal origin
0
0
0.02
0
0
1
0.12
7,676
Socialist
legal origin
0
0
0.10
0
0
1
0.31
50,982
This table reports summary statistics for aircraft age, technological age (broad and narrow), and country characteristics. The summary
statistics for aircraft age are reported for the periods 1978-1979, 1980-1989, 1990-1999, 2000-2003, as well as for the entire period. The
table also reports summary statistics separately for commercial and military aircraft.
31
Table 2:
Creditor Rights and Aircraft Vintage
Dependent
Variable=
GDP
GDP growth
GDP per capita
GDP per
capita growth
Population
Area
Creditor rights
Fixed-Effects
Year
Country
Operator
# of Countries
# of Operators
Adjusted R2
Observations
Age
Age
Age
Tech Age
(broad)
Tech Age
(broad)
Tech Age
(broad)
Tech Age
(narrow)
Tech Age
(narrow)
Tech Age
(narrow)
0.021
(0.150)
-0.056
(0.061)
-0.476 c
(0.242)
0.095 b
(0.048)
-0.531 b
(0.258)
0.362
(0.211)
-0.446 b
(0.218)
0.125
(0.162)
0.090 a
(0.032)
-2.245 a
(0.571)
0.053 a
(0.020)
10.610 a
(2.005)
6.993
(5.497)
-0.791 b
(0.339)
0.094
(0.142)
0.056 c
(0.031)
-1.444 c
(0.862)
0.050 a
(0.020)
2.959 b
(1.352)
-0.483
(1.317)
-0.668 b
(0.305)
0.007
(0.144)
0.005
(0.048)
-0.806 a
(0.284)
0.025
(0.027)
-0.647 c
(0.330)
0.398
(0.245)
-0.395 c
(0.232)
0.124
(0.142)
0.073 a
(0.031)
-1.890 a
(0.583)
0.039 a
(0.016)
8.366 a
(2.208)
-17.035 a
(4.659)
-0.696 b
(0.302)
0.075
(0.156)
0.050 c
(0.027)
-1.263 c
(0.710)
0.031 b
(0.014)
1.567
(1.279)
0.530
(1.082)
-0.443 c
(0.257)
-0.039
(0.122)
-0.038
(0.045)
-0.917 a
(0.212)
0.069 b
(0.029)
-0.511 c
(0.263)
0.413 c
(0.225)
-0.548 b
(0.229)
0.037
(0.134)
0.073 a
(0.027)
-2.719 a
(0.581)
0.042 a
(0.014)
7.851 a
(2.092)
-2.330
(4.477)
-0.527 c
(0.301)
0.054
(0.138)
0.049 b
(0.024)
-1.681 b
(0.774)
0.037 a
(0.012)
1.776
(1.369)
0.303
(1.234)
-0.301
(0.285)
Yes
No
No
129
5,883
0.09
489,407
Yes
Yes
No
129
5,883
0.12
489,407
Yes
No
Yes
129
5,883
0.45
489,407
Yes
No
No
129
5,883
0.11
489,407
Yes
Yes
No
129
5,883
0.15
489,407
Yes
No
Yes
129
5,883
0.51
489,407
Yes
No
No
129
5,883
0.15
489,407
Yes
Yes
No
129
5,883
0.18
489,407
Yes
No
Yes
129
5,883
0.51
489,407
The dependent variable is aircraft age (columns 1-3), broad technological age (columns 4-6), and narrow technological age (columns 7-9).
GDP is the natural logarithm of real GDP, GDP growth is the annual growth rate of GDP, GDP per capita is the natural logarithm of
real GDP per capita, GDP per capita growth is the annual growth rate of GDP per capita. Population is the natural logarithm of the
population. Area is the natural logarithm of the country surface in sq. km. Creditor rights is an index aggregating creditor rights, following
Djankov et al. (2007). The index ranges from 0 (weak creditor rights) to 4 (strong creditor rights) and is constructed as of January for
every year from 1978 to 2003. Columns without country or operator fixed effects also include dummies for French legal origin, German
legal origin, Nordic legal origin, and Socialist legal origin (not reported for brevity). All regressions include an intercept (not reported) and
year fixed effects. Standard-errors are clustered by country and reported in parentheses. a, b and c denote statistical significance at the
1%, 5%, and 10% levels, respectively.
32
Table 3:
Creditor Rights and Aircraft Age: Commercial Vs. Military Aircraft
GDP
GDP growth
GDP per capita
GDP per capita growth
Population
Area
Airforce size
Creditor rights
Fixed-Effects
Year
Country
Operator
# of Countries
# of Operators
Adjusted R2
Observations
Commercial
Age
Military
Age
Commercial
Age
Military
Age
Commercial
Age
Military
Age
0.180
(0.387)
-0.059
(0.053)
-0.756 c
(0.387)
0.059 b
(0.029)
-0.919 a
(0.289)
0.546 b
(0.255)
-0.177
(0.389)
-0.512 b
(0.240)
-0.153
(0.147)
-0.206
(0.164)
1.464 a
(0.353)
0.284 c
(0.166)
-0.254
(0.343)
0.071
(0.249)
-1.160 b
(0.453)
-0.351
(0.409)
0.242
(0.154)
0.092 a
(0.026)
-2.390 a
(0.623)
0.032 a
(0.011)
9.251 a
(2.361)
2.611
(5.878)
-0.774 c
(0.413)
-0.731 c
(0.412)
-0.064
(0.153)
-0.132 c
(0.068)
1.056
(1.189)
0.167 a
(0.065)
1.732
(4.402)
83.003 b
(37.051)
-5.538 a
(0.863)
-0.293
(0.597)
0.142
(0.115)
0.058 a
(0.020)
-1.815 a
(0.640)
0.030 a
(0.010)
1.800
(1.395)
0.857
(1.126)
-0.986 b
(0.412)
-0.619 b
(0.312)
-0.038
(0.173)
-0.150 b
(0.070)
1.003
(1.293)
0.178 a
(0.067)
2.381
(4.488)
-9.820
(6.390)
-5.100 a
(0.972)
-0.490
(0.646)
Yes
No
No
129
5,435
0.10
372,897
Yes
No
No
114
733
0.15
116,510
Yes
Yes
No
129
5,435
0.14
372,897
Yes
Yes
No
114
733
0.20
116,510
Yes
No
Yes
129
5,435
0.49
372,897
Yes
No
Yes
114
733
0.32
116,510
The dependent variable is aircraft age of either commercial or military aircraft. GDP is the natural logarithm of real GDP,
GDP growth is the annual growth rate of GDP, GDP per capita is the natural logarithm of real GDP per capita, GDP per
capita growth is the annual growth rate of GDP per capita. Population is the natural logarithm of the population. Area is
the natural logarithm of the country surface in sq. km. Airforce size is the fraction of military aircraft as a percentage of all
aircraft in a country. Creditor rights is an index aggregating creditor rights, following Djankov et al. (2007). The index ranges
from 0 (weak creditor rights) to 4 (strong creditor rights) and is constructed as of January for every year from 1978 to 2003.
The first two columns also include dummies for French legal origin, German legal origin, Nordic legal origin, and Socialist legal
origin (not reported for brevity). All regressions include an intercept (not reported) and year fixed effects. Standard-errors
are clustered by country and reported in parentheses. a, b and c denote statistical significance at the 1%, 5%, and 10% levels,
respectively.
33
Table 4:
Creditor Rights and Technological Age: Commercial Vs. Military Aircraft
GDP
GDP growth
GDP per capita
GDP per
capita growth
Population
Area
Airforce size
Creditor rights
Fixed-Effects
Year
Country
Operator
# of Countries
# of Operators
Adjusted R2
Observations
Commercial
Tech Age
(broad)
Military
Tech Age
(broad)
Commercial
Tech Age
(broad)
Military
Tech Age
(broad)
Commercial
Tech Age
(narrow)
Military
Tech Age
(narrow)
Commercial
Tech Age
(narrow)
Military
Tech Age
(narrow)
0.246 c
(0.140)
0.076 a
(0.024)
-2.066 b
(0.893)
0.022 b
(0.009)
8.562 a
(2.562)
-5.600
(5.573)
-0.481 b
(0.532)
-0.804 b
(0.371)
-0.040
(0.150)
-0.097 a
(0.033)
0.422
(1.086)
0.102 a
(0.031)
-5.885 c
(3.110)
40.959 c
(22.716)
-5.588 a
(0.846)
0.487
(0.639)
0.093
(0.125)
0.044 a
(0.018)
-1.423 b
(0.626)
0.016 c
(0.009)
0.659
(1.309)
1.110
(1.005)
-0.627
(0.493)
-0.680 a
(0.260)
-0.046
(0.163)
-0.114 a
(0.030)
0.380
(1.202)
0.110 a
(0.025)
-5.480 c
(3.006)
3.170
(3.537)
-5.309 a
(0.908)
0.401
(0.706)
0.169
(0.145)
0.069 a
(0.022)
-2.664 a
(0.725)
0.025 a
(0.007)
7.231 a
(2.436)
-1.877
(4.482)
-0.840 b
(0.410)
-0.678 c
(0.381)
-0.176 c
(0.103)
-0.105 a
(0.035)
0.047
(1.061)
0.125 a
(0.029)
-3.058
(2.878)
111.283 a
(20.545)
-5.047 a
(0.745)
0.629
(0.557)
0.095
(0.121)
0.041 a
(0.016)
-1.874 a
(0.605)
0.021 a
(0.006)
0.974
(1.369)
1.355
(1.078)
-0.992 b
(0.395)
-0.481 c
(0.280)
-0.141
(0.125)
-0.115 a
(0.033)
0.131
(1.231)
0.127 a
(0.027)
-2.490
(2.993)
-3.474
(3.470)
-4.619 a
(0.843)
0.533
(0.632)
Yes
Yes
No
129
5,437
0.18
372,897
Yes
Yes
No
114
893
0.31
116,510
Yes
No
Yes
129
5,437
0.48
372,897
Yes
No
Yes
114
893
0.49
116,510
Yes
Yes
No
129
5,437
0.23
372,897
Yes
Yes
No
114
893
0.26
116,510
Yes
No
Yes
129
5,437
0.53
372,897
Yes
No
Yes
114
893
0.41
116,510
The dependent variable is the broad technological age (columns 1-4), or narrow technological age (columns 5-8) of either commercial or
military aircraft. GDP is the natural logarithm of real GDP, GDP growth is the annual growth rate of GDP, GDP per capita is the
natural logarithm of real GDP per capita, GDP per capita growth is the annual growth rate of GDP per capita. Population is the natural
logarithm of the population. Area is the natural logarithm of the country surface in sq. km. Airforce size is the fraction of military
aircraft as a percentage of all aircraft in a country. Creditor rights is an index aggregating creditor rights, following Djankov et al. (2007).
The index ranges from 0 (weak creditor rights) to 4 (strong creditor rights) and is constructed as of January for every year from 1978 to
2003. All regressions include an intercept (not reported) and year fixed effects. Standard-errors are clustered by country and reported in
parentheses. a, b and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
34
Table 5:
Creditor Rights and Age: Leased Vs. Non-Leased Aircraft
GDP
GDP growth
GDP per capita
GDP per
capita growth
Population
Area
Creditor rights
Fixed-Effects
Year
Country
Operator
# of Countries
# of Operators
Adjusted R2
Observations
Non-Leased
Age
Leased
Age
Non-Leased
Age
Leased
Age
Non-Leased
Age
Leased
Age
Non-Leased
Tech Age
(broad)
Leased
Tech Age
(broad)
0.408
(0.271)
-0.245 b
(0.113)
-1.159 a
(0.374)
0.168
(0.230)
-1.373 a
(0.487)
0.938 a
(0.287)
-0.984 a
(0.376)
-0.111
(0.137)
-0.159 a
(0.057)
-0.778 b
(0.320)
0.029
(0.018)
-1.026 a
(0.355)
0.926
(0.202)
-0.155
(0.243)
0.515 c
(0.278)
0.085 b
(0.036)
-3.614 a
(0.747)
0.067 a
(0.025)
9.949 a
(2.634)
42.389 a
(16.105)
-0.699
(0.475)
0.031
(0.146)
0.094 a
(0.028)
-2.129 a
(0.587)
0.016 a
(0.005)
8.020 b
(3.870)
10.853 b
(5.161)
-0.320
(0.450)
0.331 c
(0.180)
0.048
(0.036)
-2.909 a
(0.970)
0.063 b
(0.029)
5.106 b
(2.304)
-4.013 b
(1.719)
-0.651 b
(0.323)
-0.115
(0.136)
0.069 a
(0.026)
-2.322
(0.531)
0.016 a
(0.002)
2.169 c
(1.288)
1.287 c
(0.701)
-0.237
(0.418)
0.315
(0.287)
-0.231 b
(0.112)
-1.470 a
(0.372)
0.140
(0.119)
-1.291 a
(0.466)
0.866 a
(0.312)
-0.732 b
(0.357)
-0.026
(0.124)
-0.114 b
(0.061)
-1.332 a
(0.273)
0.021
(0.019)
-0.789 c
(0.417)
0.871 a
(0.271)
0.132
(0.226)
Yes
No
No
128
3,750
0.11
213,250
Yes
No
No
129
3,600
0.05
159,647
Yes
Yes
No
128
3,750
0.20
213,250
Yes
Yes
No
129
3,600
0.11
159,647
Yes
No
Yes
128
3,750
0.52
213,250
Yes
No
Yes
129
3,600
0.53
159,647
Yes
No
No
128
3,750
0.17
213,250
Yes
No
No
129
3,600
0.07
159,647
The dependent variable is aircraft age or broad technological age of either non-leased or leased aircraft. GDP is the natural logarithm of
real GDP, GDP growth is the annual growth rate of GDP, GDP per capita is the natural logarithm of real GDP per capita, GDP per
capita growth is the annual growth rate of GDP per capita. Population is the natural logarithm of the population. Area is the natural
logarithm of the country surface in sq. km. Creditor rights is an index aggregating creditor rights, following Djankov et al. (2007). The
index ranges from 0 (weak creditor rights) to 4 (strong creditor rights) and is constructed as of January for every year from 1978 to
2003. Columns without country or operator fixed effects also include dummies for French legal origin, German legal origin, Nordic legal
origin, and Socialist legal origin (not reported for brevity). All regressions include an intercept (not reported) and year fixed effects.
Standard-errors are clustered by country and reported in parentheses. a, b and c denote statistical significance at the 1%, 5%, and 10%
levels, respectively.
35
Table 6:
Creditor Rights and Age of Non-Leased Aircraft: Poor Vs. Rich Countries
GDP
GDP growth
GDP per capita
GDP per
capita growth
Population
Area
Creditor rights
Fixed-Effects
Year
Country
Operator
# of Countries
# of Operators
Adjusted R2
Observations
Rich
countries
Poor
countries
Rich
countries
Poor
countries
Rich
countries
Poor
countries
Rich countries (U.S Excluded)
Age
Age
Age
Age
Age
Age
Age
Age
Age
0.483 c
(0.274)
-0.033 b
(0.091)
-1.805 a
(0.329)
0.106
(0.087)
-0.453 a
(0.447)
0.762 a
(0.252)
-0.715 b
(0.355)
0.351 a
(0.132)
-0.136
(0.264)
-1.553
(1.037)
0.042
(0.286)
-2.118 a
(0.515)
0.117
(0.522)
0.172
(0.460)
0.655 a
(0.250)
0.085 b
(0.039)
-4.579 a
(1.620)
0.064 a
(0.024)
9.023 a
(2.896)
47.645 a
(14.562)
-1.017 a
(0.335)
0.014
(0.112)
0.268 c
(0.159)
-3.380 a
(0.735)
-0.186
(0.162)
0.370
(6.180)
-731.985 b
(336.94)
-0.025
(0.634)
0.368 c
(0.189)
0.063
(0.039)
-5.005 a
(1.677)
0.056 b
(0.027)
2.743
(2.031)
-2.274
(1.418)
-0.648 b
(0.311)
0.148
(0.115)
0.220
(0.142)
-2.695 a
(0.427)
-0.168
(0.143)
-0.869
(4.260)
-1.904 c
(3.924)
0.345
(0.548)
-0.519 c
(0.291)
-0.097
(0.089)
-1.040 b
(0.429)
0.135
(0.093)
0.308
(0.527)
0.729 a
(0.256)
-0.607 c
(0.339)
-0.233
(0.316)
0.061 c
(0.034 )
-4.354 a
(1.534)
0.066 b
(0.026)
10.043 a
(2.949)
44.943 a
(16.585)
-0.941 a
(0.348)
-0.294
(0.299)
0.051
(0.038)
-4.668 a
(1.489)
0.059 b
(0.029)
5.288 b
(2.609)
-5.433 c
(2.926)
-0.686 b
(0.293)
Yes
No
No
66
2,963
0.12
177,937
Yes
No
No
72
845
0.26
35,313
Yes
Yes
No
66
2,963
0.16
177,937
Yes
Yes
No
72
845
0.37
35,313
Yes
No
Yes
66
2,963
0.51
177,937
Yes
No
Yes
72
845
0.58
35,313
Yes
No
No
65
2,093
0.17
106,027
Yes
Yes
No
65
2,093
0.25
106,027
Yes
No
Yes
65
2,093
0.54
106,027
The dependent variable is aircraft age in either rich or poor countries. Rich (poor) countries are defined in each year as those with GDP
per capita greater (smaller) than the sample median of that year. GDP is the natural logarithm of real GDP, GDP growth is the annual
growth rate of GDP, GDP per capita is the natural logarithm of real GDP per capita, GDP per capita growth is the annual growth rate
of GDP per capita. Population is the natural logarithm of the population. Area is the natural logarithm of the country surface in sq. km.
Creditor rights is an index aggregating creditor rights, following Djankov et al. (2007). The index ranges from 0 (weak creditor rights) to
4 (strong creditor rights) and is constructed as of January for every year from 1978 to 2003. Columns without country or operator fixed
effects also include dummies for French legal origin, German legal origin, Nordic legal origin, and Socialist legal origin (not reported for
brevity). All regressions include an intercept (not reported) and year fixed effects. Standard-errors are clustered by country and reported
in parentheses. a, b and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
36
Table 7:
Creditor Rights Government Ownership and Aircraft Vintage
GDP
GDP growth
GDP per capita
GDP per
capita growth
Population
Area
Government
Creditor rights
Fixed-Effects
Year
Country
Operator
# of Countries
# of Operators
Adjusted R2
Observations
Non-Leased
Age
Leased
Age
Non-Leased
Age
Leased
Age
Non-Leased
Age
Leased
Age
Non-Leased
Tech Age
(broad)
Leased
Tech Age
(broad)
0.384
(0.270)
-0.203 b
(0.102)
-1.326 a
(0.386)
0.128
(0.101)
-1.398 a
(0.495)
0.824 a
(0.287)
-2.760 a
(0.431)
-1.067 a
(0.357)
-0.122
(0.131)
-0.144 b
(0.055)
-0.896 b
(0.324)
0.029 c
(0.016)
-1.074 a
(0.340)
0.844 a
(0.195)
-2.676 a
(0.826)
-0.297
(0.236)
0.511 c
(0.274)
0.084 b
(0.037)
-3.356 a
(0.722)
0.064 a
(0.024)
10.134 a
(2.780)
28.776
(20.156)
-2.797 a
(0.390)
-0.704
(0.490)
0.021
(0.143)
0.097 a
(0.028)
-2.103 a
(0.583)
0.015 a
(0.004)
7.614 c
(3.959)
3.363
(7.745)
-1.792 b
(0.895)
-0.282
(0.440)
0.330 c
(0.178)
0.048
(0.036)
-2.904 a
(0.958)
0.063 b
(0.029)
4.790 b
(2.226)
-3.785 b
(1.646)
1.507
(0.973)
-0.619 b
(0.298)
-0.113
(0.138)
0.069 a
(0.026)
-2.322
(0.530)
0.016 a
(0.002)
2.174 c
(1.288)
1.285 c
(0.701)
0.571
(0.833)
-0.249
(0.404)
0.294
(0.286)
-0.196 c
(0.104)
-1.611 a
(0.377)
0.106
(0.105)
-1.312 a
(0.469)
0.771 b
(0.303)
-2.304 a
(0.430)
-0.801 b
(0.339)
-0.036
(0.120)
-0.139 b
(0.058)
-1.443 a
(0.280)
0.021
(0.018)
-0.833 b
(0.407)
0.794 a
(0.260)
-2.495 a
(0.925)
0.0001
(0.218)
Yes
No
No
128
3,750
0.13
213,250
Yes
No
No
129
3,600
0.06
159,647
Yes
Yes
No
128
3,750
0.20
213,250
Yes
Yes
No
129
3,600
0.11
159,647
Yes
No
Yes
128
3,750
0.52
213,250
Yes
No
Yes
129
3,600
0.53
159,647
Yes
No
No
128
3,750
0.18
213,250
Yes
No
No
129
3,600
0.07
159,647
The dependent variable is aircraft age or broad technological age in either non-leased or leased aircraft. GDP is the natural logarithm of
real GDP, GDP growth is the annual growth rate of GDP, GDP per capita is the natural logarithm of real GDP per capita, GDP per
capita growth is the annual growth rate of GDP per capita. Population is the natural logarithm of the population. Area is the natural
logarithm of the country surface in sq. km. Government is a dummy variable that equals one if the airline operating the aircraft is at
least partially owned by the government and equals zero otherwise. Creditor rights is an index aggregating creditor rights, following
Djankov et al. (2007). The index ranges from 0 (weak creditor rights) to 4 (strong creditor rights) and is constructed as of January for
every year from 1978 to 2003. Columns without country or operator fixed effects also include dummies for French legal origin, German
legal origin, Nordic legal origin, and Socialist legal origin (not reported for brevity). All regressions include an intercept (not reported)
and year fixed effects. Standard-errors are clustered by country and reported in parentheses. a, b and c denote statistical significance at
the 1%, 5%, and 10% levels, respectively.
37
Table 8:
Aircraft Age and Aviation Regulation
GDP
GDP growth
GDP per capita
GDP per
capita growth
Population
Area
Government
Aviation
Creditor rights
Commercial
Leased
and
Non-Leased
Commercial
Commercial
Military
Military
and
Commercial
All
Military
and
Commercial
Non-Leased
Commercial
Leased
and
Non-Leased
Non-Leased
Leased
Age
Age
Age
Age
Age
Age
Age
-0.026
(0.256)
-0.170
(0.108)
-1.134 a
(0.506)
0.045
(0.033)
-1.567 a
(0.467)
1.353 a
(0.313)
-3.854 a
(0.744)
-3.277 a
(1.079)
-0.820 b
(0.388)
0.255
(0.425)
-0.133
(0.139)
-1.396 b
(0.605)
0.049
(0.070)
-1.915 a
(0.623)
1.310 a
(0.342)
-4.470 a
(0.783)
-3.858 a
(1.138)
-1.844 a
(0.522)
-0.161
(0.176)
-0.209 a
(0.071)
-0.640
(0.443)
0.035 c
(0.019)
-1.114 a
(0.321)
1.016 a
(0.236)
-3.564 a
(1.001)
-2.983 b
(1.194)
-0.290
(0.274)
-0.175
(0.173)
-0.091
(0.099)
0.067
(0.386)
0.150
(0.095)
-1.157 a
(0.381)
0.745 a
(0.282)
-0.094
(0.240)
-0.113
(0.087)
-0.844 b
(0.414)
0.055
(0.037)
-1.468 a
(0.403)
1.244 a
(0.261)
-3.627 a
(0.698)
-2.453 a
(0.926)
-0.800 b
(0.356)
4.122 a
(1.334)
0.601
(0.958)
0.088
(0.321)
-0.100
(0.095)
-0.915 b
(0.449)
0.081
(0.066)
-1.683 a
(0.468)
1.168 a
(0.260)
-4.067 a
(0.757)
-2.307 b
(0.993)
-1.780 a
(0.465)
0.995
(1.224)
1.604 c
(0.843)
0.030
(0.272)
-0.162
(0.108)
-1.039 b
(0.504)
0.039
(0.028)
-1.522 a
(0.451)
1.185 a
(0.285)
-4.119 a
(0.691)
-3.397 a
(1.083)
-1.837 a
(0.461)
0.745
(1.58)
0.171
(0.829)
Military
Military×
Creditor rights
Leased
-5.444 a
(0.439)
1.556 a
(0.246)
Leased×
Creditor rights
Fixed-Effects
Year
# of Countries
# of Operators
Adjusted R2
Observations
Yes
67
3,254
0.10
188,142
Yes
67
2,226
0.13
93,871
Yes
67
2,170
0.06
94,271
Yes
64
520
0.04
51,696
Yes
67
3,774
0.13
239,838
Yes
67
2,746
0.11
145,567
Yes
67
3,254
0.13
188,142
The dependent variable is aircraft age of commerical non-leased aircraft, commercial leased aircraft, or military aircraft. GDP is the
natural logarithm of real GDP, GDP growth is the annual growth rate of GDP, GDP per capita is the natural logarithm of real GDP
per capita, GDP per capita growth is the annual growth rate of GDP per capita. Population is the natural logarithm of the population.
Area is the natural logarithm of the country surface in sq. km. Government is a dummy variable that equals one if the airline operating
the aircraft is at least partially owned by the government and equals zero otherwise. Aviation is a dummy variable that equals one for
aircraft in countries that comply with ICAO standards and equals zero otherwise. Military is a dummy variable that equals one for
military aircraft and zero otherwise. Leased is a dummy variable that equals one for leased aircraft and zero otherwise. Creditor rights
is an index aggregating creditor rights, following Djankov et al. (2007). The index ranges from 0 (weak creditor rights) to 4 (strong
creditor rights) and is constructed as of January for every year from 1978 to 2003. All regressions include an intercept (not reported),
legal origins dummies, and year fixed effects. Standard-errors
38are clustered by country and reported in parentheses. a, b and c denote
statistical significance at the 1%, 5%, and 10% levels, respectively.
Table 9:
Creditor Rights, Leverage and Aircraft Age: Leased vs. Non-Leased Aircraft
GDP
GDP growth
GDP per capita
GDP per
capita growth
Population
Area
Government
Aviation
Creditor rights
Leverage
Creditor rights
×Leverage
Non-Leased
Leased
Non-Leased
Leased
Non-Leased
Leased
Non-Leased
Leased
Age
Age
Age
Age
Age
Age
Age
Age
0.982 b
(0.437)
0.102
(0.108)
0.025
(0.839)
-0.023
(0.016)
-0.119
(0.332)
0.399
(0.255)
-1.507
(1.378)
-8.643
(3.240)
0.426
(0.686)
6.581 b
(3.224)
-3.158 b
(1.378)
-0.328 c
(0.179)
-0.023
(0.036)
0.994 a
(0.285)
0.020 a
(0.005)
0.614 b
(0.264)
0.023
(0.206)
0.123
(0.556)
-6.415 a
(1.548)
-0.029
(0.572)
4.330 a
(1.546)
-0.279
(0.946)
0.839 c
(0.440)
164 c
(0.084)
0.022
(2.366)
-0.020
(0.029)
-14.966
(17.276)
7.281
(63.774)
1.537
(1.042)
4.474 a
(1.399)
1.342
(0.793)
9.038 a
(2.923)
-5.067 a
(1.626)
-0.261
(0.213)
-0.011
(0.036)
-1.404
(2.154)
0.020 a
(0.003)
9.332
(9.441)
-102.413 a
(5.306)
1.421
(1.232)
1.614 a
(0.419)
0.275
(0.854)
3.947 b
(1.538)
0.204
(1.020)
0.942 a
(0.431)
0.089
(0.114)
-0.016
(0.836)
-0.023
(0.015)
-0.085
(0.310)
0.404
(0.258)
-1.564
(1.444)
-8.969 b
(3.303)
0.315
(0.783)
-0.396 b
(0.184)
-0.025
(0.036)
0.944 a
(0.294)
0.021 a
(0.006)
0.654 b
(0.251)
0.046
(0.203)
-0.020
(0.571)
-6.880 a
(1.569)
-0.115
(0.511)
0.852 c
(0.432)
0.171 c
(0.087)
-0.269
(2.339)
-0.023
(0.030)
-16.029
(16.878)
4.121
(62.968)
1.594
(1.068)
4.606 a
(1.467)
0.910
(0.805)
-0.325
(0.220)
0.008
(0.038)
-1.203
(2.255)
0.019 a
(0.003)
9.234
(9.266)
-97.970 a
(5.591)
1.420
(1.258)
1.630 a
(0.428)
0.104
(0.789)
5.719
(4.381)
-3.212 c
(1.782)
3.786 a
(1.350)
-0.116
(0.899)
9.412 b
(3.648)
-5.737 b
(2.178)
3.140 b
(1.399)
0.510
(1.007)
Yes
No
28
0.11
33,652
Yes
No
28
0.04
29,384
Yes
Yes
28
0.13
33,652
Yes
Yes
28
0.06
29,384
LT Debt
Creditor rights
×LT Debt
Fixed-Effects
Year
Country
# of Countries
Adjusted R2
Observations
Yes
No
28
0.11
33,652
Yes
No
28
0.05
29,384
Yes
Yes
28
0.13
33,652
Yes
Yes
28
0.06
29,384
The dependent variable is aircraft age of either leased or non-leased aircraft. GDP is the natural logarithm of real GDP, GDP growth
is the annual growth rate of GDP, GDP per capita is the natural logarithm of real GDP per capita, GDP per capita growth is the
annual growth rate of GDP per capita. Population is the natural logarithm of the population. Area is the natural logarithm of the
country surface in sq. km. Government is a dummy variable that equals one if the airline operating the aircraft is at least partially
owned by the government and equals zero otherwise. Aviation is a dummy variable that equals one for aircraft in countries that
comply with ICAO standards and equals zero otherwise. Creditor rights is an index aggregating creditor rights, following Djankov et
al. (2007). The index ranges from 0 (weak creditor rights) to 4 (strong creditor rights) and is constructed as of January for every year
from 1978 to 2003. Leverage is total debt divided by total assets. LT Debt is long-term debt divided by total assets. Columns 1,2,5
and 6 also include dummies for French legal origin, German legal origin, Nordic legal origin, and Socialist legal origin (not reported
for brevity). All regressions include an intercept (not reported) and year fixed effects. Standard-errors are clustered by country and
reported in parentheses. a, b and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
39
Table 10:
Creditor Rights and Fleet Size: Commercial Vs. Military Aircraft
Size=
GDP per capita
GDP growth
GDP per capita
GDP per
capita growth
Population
Area
Government
Creditor rights
Fixed-Effects
Year
Operator
# of Countries
# of Operators
Adjusted R2
Observations
Commercial
Number
Military
Number
Commercial
Seats
Military
Seats
Commercial
Weight
Military
Weight
Commercial
Wings
Military
Wings
0.024 b
(0.011)
-0.006
(0.004)
0.227 a
(0.056)
-0.001
(0.003)
0.047
(0.133)
-0.106
(0.105)
-0.007
(0.027)
0.055 b
(0.027)
0.022
(0.097)
-0.006 a
(0.036)
0.328 b
(0.129)
0.002 b
(0.001)
0.047
(0.213)
0.360
(0.286)
0.116 b
(0.052)
0.001
(0.005)
0.597
(0.477)
-0.007 b
(0.003)
0.113
(0.763)
0.411
(0.923)
-0.029
(0.126)
0.061 b
(0.025)
-0.012
(0.009)
0.508 a
(0.129)
-0.001
(0.006)
0.096
(0.298)
-0.184
(0.251)
0.077
(0.234)
0.138 b
(0.059)
0.073
(0.044
-0.014 a
(0.005)
0.775 b
(0.309)
0.004 c
(0.002)
0.036
(0.574)
0.746
(0.726)
0.009
(0.118)
0.088 b
(0.038)
-0.011
(0.010)
0.611 a
(0.166)
0.007 a
(0.006)
0.244
(0.337)
-0.104
(0.297)
0.118
(0.214)
0.183 b
(0.091)
0.070
(0.055)
-0.047 b
(0.006)
0.921 b
(0.439)
0.005 c
(0.003)
-0.323
(0.921)
1.354
(1.091)
-0.061
(0.049)
0.060 b
(0.026)
-0.011
(0.020)
0.616 a
(0.154)
-0.0003
(0.007)
0.217
(0.324)
-0.202
(0.271)
0.310
(0.218)
0.182 a
(0.062)
Yes
Yes
129
5,284
0.85
32,913
Yes
Yes
110
616
0.93
6,164
Yes
Yes
129
5,284
0.87
32,913
Yes
Yes
110
616
0.91
6,164
Yes
Yes
129
5,284
0.83
32,913
Yes
Yes
110
616
0.92
6,164
Yes
Yes
129
5,284
0.86
32,913
Yes
Yes
110
616
0.94
6,164
-0.058
(0.101)
The dependent variable is fleet size defined as the logarithm of either 1) the sum of all aircraft operated, (2) the sum of the seat
capacities of all aircraft in the fleet, (3) the sum of the maximal takeoff weight of all aircraft in the fleet, and (4) the sum of the
wingspans of all aircraft in the fleet. GDP is the natural logarithm of real GDP, GDP growth is the annual growth rate of GDP, GDP
per capita is the natural logarithm of real GDP per capita, GDP per capita growth is the annual growth rate of GDP per capita.
Population is the natural logarithm of the population. Area is the natural logarithm of the country surface in sq. km. Government is
a dummy variable that equals one if the airline operating the aircraft is at least partially owned by the government and equals zero
otherwise. Creditor rights is an index aggregating creditor rights, following Djankov et al. (2007). The index ranges from 0 (weak
creditor rights) to 4 (strong creditor rights) and is constructed as of January for every year from 1978 to 2003. All regressions include
an intercept (not reported), year fixed effects and operator fixed-effects. Standard-errors are clustered by country and reported in
parentheses. a, b and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
40
Document related concepts
no text concepts found